Table 1.
Pairwise comparisons of treatment group effects on the risk outcomes
| Glargine | Glimepiride | Liraglutide | Sitagliptin | Total | |
|---|---|---|---|---|---|
| Composite outcomea (P < 0.001) | |||||
| No. of participantse | 1,250 | 1,237 | 1,242 | 1,255 | 4,984 |
| No. who reached outcome (%) | 899 (71.92) | 1,007 (81.41) | 749 (60.31) | 874 (69.64) | 3,529 (70.81) |
| Crude rate (SE), n/100 years | 29.0 (1.0) | 39.9 (1.3) | 19.4 (0.7) | 26.2 (0.9) | 27.5 (0.5) |
| Pairwise HR (CI)d | |||||
| Glimepiride | 0.74 (0.68, 0.81)*** | ||||
| Liraglutide | 1.49 (1.35, 1.63)*** | 2.00 (1.82, 2.19)*** | |||
| Sitagliptin | 1.11 (1.01, 1.22)* | 1.50 (1.37, 1.64)*** | 0.75 (0.68, 0.82)*** | ||
| Sensitivity analysisb | |||||
| Pairwise HR (CI)d | |||||
| Glimepiride | 0.77 (0.70, 0.85)*** | ||||
| Liraglutide | 1.54 (1.37, 1.73)*** | 1.99 (1.77, 2.24)*** | |||
| Sitagliptin | 1.17 (1.08, 1.27)*** | 1.51 (1.36, 1.69)*** | 0.76 (0.68, 0.85)*** | ||
| HbA1c outcome: >7.5% or 58 mmol/mol, confirmed (P < 0.001) | |||||
| No. of participants | 1,263 | 1,254 | 1,262 | 1,268 | 5,047 |
| No. who reached outcome (%) | 498 (39.43) | 633 (50.48) | 583 (46.2) | 697 (54.97) | 2,411 (47.77) |
| Crude rate (SE), n/100 years | 10.7 (0.5) | 14.8 (0.6) | 13.0 (0.5) | 17.5 (0.7) | 13.9 (0.3) |
| Pairwise HR (CI)d | |||||
| Glimepiride | 0.73 (0.64, 0.82)*** | ||||
| Liraglutide | 0.83 (0.73, 0.93)** | 1.13 (1.01, 1.27)* | |||
| Sitagliptin | 0.61 (0.54, 0.69)*** | 0.84 (0.75, 0.93)** | 0.74 (0.66, 0.83)*** | ||
| Weight outcome: weight gain ≥5% (P < 0.001) | |||||
| No. of participantse | 1,246 | 1,239 | 1,240 | 1,255 | 4,980 |
| No. who reached outcome (%) | 552 (44.3) | 531 (42.86) | 257 (20.73) | 390 (31.08) | 1,730 (34.7) |
| Crude rate (SE), n/100 years | 13.5 (0.6) | 13.0 (0.6) | 4.9 (0.3) | 8.1 (0.4) | 9.5 (0.2) |
| Pairwise HR (CI)d | |||||
| Glimepiride | 1.03 (0.91, 1.16) | ||||
| Liraglutide | 2.68 (2.33, 3.08)*** | 2.61 (2.3, 3.0)*** | |||
| Sitagliptin | 1.63 (1.44, 1.85)*** | 1.59 (1.4, 1.8)*** | 0.61 (0.5, 0.7)*** | ||
| Hypoglycemiac outcome (P < 0.001) | |||||
| No. of participantse | 1,245 | 1,231 | 1,233 | 1,253 | 4,962 |
| No. who reached outcome (%) | 474 (38.1) | 654 (53.1) | 312 (25.3) | 328 (26.2) | 1,768 (35.8) |
| Crude rate (SE), n/100 years | 9.6 (0.4) | 15.3 (0.6) | 5.8 (0.3) | 6.0 (0.3) | 8.8 (0.2) |
| Pairwise HR (CI)d | |||||
| Glimepiride | 0.62 (0.55, 0.70)*** | ||||
| Liraglutide | 1.64 (1.43, 1.89)*** | 2.64 (2.32, 3.00)*** | |||
| Sitagliptin | 1.59 (1.39, 1.82)*** | 2.56 (2.32, 3.00)*** | 0.97 (0.84, 1.12) |
Shown are the number of participants at risk and number of cases, event rates, and pairwise comparisons of treatment group effects (columns vs. rows) on the risk of the composite outcome and individual components during study follow-up (mean study duration 5 years).
aTime to the earliest of the following three outcomes: HbA1c >7.5% confirmed, weight gain from baseline ≥5%, adjudicated SH or patient-reported nonsevere hypoglycemia.
bWei-Lachin analysis with reported HRs computed from the average of the log HRs for the three components of the composite outcome. Details of the analysis can be found in Statistical Analysis (Research Design and Methods).
cAny SH or two or more symptomatic episodes within a short period of time (i.e., within 30 days prior to each clinic visit).
dThe pairwise HRs comparing groups (columns vs. rows) were obtained from Cox PH models for time to first occurrence of that particular outcome; e.g., HR 0.74 (95% CI 0.68, 0.81) for the comparison of glargine vs. glimepiride groups on risk of the composite outcome and 2.00 (1.82, 2.19) for the comparison of glimepiride vs. liraglutide groups on risk of the composite outcome. The P values for pairwise comparisons are reported with asterisks:
*0.01 < P ≤ 0.05;
**0.001 < P ≤ 0.01;
***P ≤ 0.001. Note that P values and 95% CIs are adjusted for the six pairwise treatment comparisons for each outcome.
eThere were 85 (1.68%) participants who either had prevalent hypoglycemia at baseline or who could not be assessed for hypoglycemia during follow-up because of missing data (glargine, 18; glimepiride, 23; liraglutide, 29; and sitagliptin, 15). There were 67 (1.33%) participants who could not be assessed for weight gain during follow-up because of missing data (glargine, 17; glimepiride, 15; liraglutide, 22; sitagliptin, 13). There were 63 (1.25%) participants excluded from the composite outcome analyses due to missing hypoglycemia and weight change outcomes described above.